Study On Reflection Spectral Characteristics And Content Inversion Of Manganese Or Uranium Contaminated Soil

The improper exploitation of manganese and uranium resources causes soil pollution and endangers human health.Therefore,it is important to monitor the soil contaminated by manganese and uranium.Traditional monitoring methods are mainly chemical tests,which are time consuming,while hyperspectral remote sensing improves this situation.Add manganese metal(concentration 800-9600mg/kg)and uranium(concentration 25-800mg/kg)to uncontaminated soil samples,sample after several months of sequestration,and obtain soil reflectance spectrum by ground spectral spectrometer PSR-2500.After spectral smoothing,the differential information and the inverse logarithmic transformation method are used to highlight the effective information of the soil spectrum and obtain the spectral characteristic parameters.Using the Pearson method,the correlation coefficient between soil spectrum and manganese and uranium content was calculated.The characteristic bands of soil spectra and soil characteristic parameters were selected as explanatory variables to perform spectral modeling inversion,and the soil spectral library was constructed.The main research results are as follows:(1)The spectral reflectance of soil contaminated by manganese is between 5% and 40%.In the 400-900 nm band,the soil reflectivity decreases with the increase of manganese concentration.When the concentration reaches 4800mg/kg,the reflectivity is the lowest,and the concentration continues to increase.The reflectivity rises.The spectral reflectance of uranium-contaminated soil is between 5% and 35%.In the 400-900 nm band,the reflectivity of soil decreases with increasing uranium concentration.When the uranium concentration is about 275 mg/kg,the spectral reflectance of soil sample is the lowest;at 900-In the 2500 nm interval,the uncontaminated soil sample has the highest reflectivity.After uranium pollution,the reflectivity decreases with the increase of concentration.When the uranium concentration is 375mg/kg,the spectral reflectance of the soil sample is the lowest.(2)After the smoothed spectrum is transformed by first-order differential transformation,the characteristic bands of manganese pollution are concentrated at 420-610 nm,660-960 nm,1032-1421 nm,2147-2230nm;the characteristic bands under uranium pollution are mostly located at 440-830 nm and infrared.Band.The second-order differential transformation of the spectrum shows that the characteristic bands of manganese pollution are distributed in the whole band;the characteristic bands of uranium pollution are distributed around 350-420 nm,1408nm and 2130 nm.The spectrum is transformed by reciprocal logarithm.The characteristic bands of manganese pollution are 380-980 nm,1635-2360 nm and 2530 nm.The characteristic bands of uranium pollution are 330-400 nm and 510-2535 nm.The spectral characteristic parameters tested by p=0.05 under manganese pollution are the soil absorption valley width,depth,area,and blue-edge peak;the parameters tested by p=0.05 under uranium pollution are the absorption valley depth,location and area.(3)Using a multiple linear regression analysis and neural network model to establish a model of the relationship between soil spectrum and soil manganese and uranium content.The results show that the best inversion model for manganese content is the neural network model after second-order differential transformation of the spectrum,with R2 of 0.966 and RMSE of 0.488.The best inversion model for uranium content is the neural network model after spectral first-order differential transformation,with R2 of 0.956 and RMSE of 0.142.(4)Establish an experimental soil spectral library,which contains soil spectral curves under different concentrations of manganese and uranium.The results of the paper can broaden the monitoring channels for soil manganese and uranium in mining areas.